ReactiveAI/TinyStories-MRL
收藏Hugging Face2025-09-10 更新2025-07-05 收录
下载链接:
https://hf-mirror.com/datasets/ReactiveAI/TinyStories-MRL
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是一个用于记忆强化学习(MRL)的合成数据集,用于训练反应式Transformer模型。数据集分为多个子集,用于MRL训练的不同阶段。每个子集包含不同数量的后续交互,可以使用不同的策略,并具有训练和验证划分。数据集基于TinyStories数据集,包括带有故事及其细节的问题/答案。数据集由Qwen3模型系列生成,并经过过滤以确保多样性和正确性。
The dataset is a synthetic Memory Reinforcement Learning (MRL) dataset for training event-driven reactive Transformer models. It is divided into subsets for different MRL curriculum stages, each with a different number of follow-up interactions and training strategies. The dataset is based on the TinyStories dataset and includes stories with question/answers. It is generated using Qwen3 models and filtered for diversity and correctness.
提供机构:
ReactiveAI



